How to make data and AI add up

In a well-worn cliché, data is often referred to as “the new oil”. The analogy is limited, but it does have some truth to it as data — like oil — is the defining resource for a new industrial age. Likewise, data seems set to be dominated by a small number of massive global players.

For organisations hoping to become pioneers in artificial intelligence (AI) and data analytics, scale confers significant competitive advantages. Bigger companies will be better placed to build the bigger data sets that enable more sophisticated analysis to be performed more quickly.

The importance of scale means that success in AI is often self-perpetuating. Amazon, for example, has taken the lead among companies developing voice assistants because of the virtuous cycle initiated by its early Alexa-enabled devices. The effectiveness of early iterations of the software, combined with compelling hardware, won it more and more users, which in turn generated exponentially more data to train and improve the assistant, attracting more users, and so on.

Financial Times Executive Education rankings 2019

Which schools are in the top 80 for Customised and Open-enrolment programmes 2019? Find out which schools are the top 50 providers for both custom and open courses. Also, learn how the tables are compiled.

Of course, not all companies can — or should, or would even want to — build a business on monetising data. For the majority, their interest in AI and data analytics will be in the application of tools to their existing business models and processes to enable reduced costs, better targeting, greater efficiency or deeper customer insight, among other benefits. They know, or will soon find out the hard way, that if they don’t capitalise on the vast amount of data up for grabs — either because they trust “old-school” intuition or because they simply don’t have the technical or human resources to handle it effectively — then someone else will, and they risk being left behind.

Managers and executives don’t need to learn the detail of data analysis methods and tools. But they do need to understand the business and strategic challenges and opportunities that those methods and tools could address, while empowering specialists who understand the technicalities to lead the work. Staying on the technical frontier is as much an organisational challenge as it is a technical one, requiring structures that enable managers and technical specialists to talk meaningfully to each other.

While some education of executives will be required, the biggest step that organisations can take towards that goal is investing in a new function: the so-called “data executive” or “data product manager”, whose role is to act as a go-between and bridge the gaps in understanding each side’s challenges, priorities and achievements in the other camp.

It is difficult to say at this point exactly what skill set these individuals should ideally have. Should they be data people primarily, with some business training, or vice versa? But their role has parallels with that of the technology consultants who emerged 20 or 30 years ago. While those consultants were helping to bridge the gap between business and IT and turn predominantly analogue organisations into digital operators, the new data executives will help businesses replace — or at least supplement — intuition with data-led decision-making.

It will be particularly important for top managers to manage the ethical quandaries around the collection and use of data. The C-suite will need to ensure that its organisation applies AI and data analytics responsibly.

Some decisions will be straightforward. I recently spoke to a manager of an energy exploration company that wanted to use data analytics to improve drilling efficiency. Once you accept fossil fuels will remain part of the energy mix for some time during the transition to renewables, this is an appropriate use of a learning algorithm.

But the manager of a large police force on a recent executive course was aware that while there was potential value in “predictive policing” — the use of analytics to pre-emptively identify potential criminal activity and actors — the area was rife with moral ambiguity needing thoughtful management. Likewise, I admit to having once been blindly confident about the potential of algorithmically driven hiring and matching of life partners. Smart managers now know that algorithms are as good as the data you train them on, retaining the human biases present in that information.

In weighing up the potential impact of AI and data analytics on their businesses, executives should focus on enabling better decision-making. What are the biggest business challenges facing their industry in general and their company specifically? What information would help them make better decisions about those challenges? Thinking a few steps ahead, executives can figure out what they need to know, then work with data scientists and engineers to find out either how to dig the answers out of the data already in the company’s possession. Or they can design and carry out experiments and other analytical initiatives that will provide them with those answers and enable them to allocate resources in an efficient manner.

Paul Oyer is the Mary and Rankine Van Anda Entrepreneurial Professor and Professor of Economics at Stanford Graduate School of Business

Letter in response to this article:

More education needed on data breaches, and privacy / From Mike Frost, Bristol, UK